Two gateway's of last resorts

[TRACY HARTMANN]

I have two circuits that I run EIGRP, the metrics perfer the bigger circuit and when it fails it uses the 2nd circuit.   The issue is the gateway of last resort.   I have two static ip route 0.0.0.0 0.0.0.0 X.X.X.X  and have one with a metric.   We had a fiber cut and the gateway of last resort did NOT move to the route with the metric.  I had to remove the original to force it to use the other circuit.

What is the best way to have this fail over to the 2nd circuit for the gateway of last resort?

[David Ruess]

Hello,

How is the static route set up? If its referencing something that remained up when the fiber was cut then it would not remove the static route, since according to it, its still functional. 

You could use IP SLA, which is a tracking mechanism that allows you to track lots of things like IPs and interface status's. You could configure it to ping a certain IP and if that becomes unreachable then the new Default route would take over.

-David

[TRACY HARTMANN]

I just have a metric on the two default gateways, I had thought about the IP SLA but wasn't sure if there is a better option.  

[David Ruess]

"Better" is subjective. IPS SLA in AN option but not the only option. Ultimately you have to determine what is "better" for your network/needs. We would possibly need more information such as remote devices configuration to see how its set up and how EIGRP is advertising networks to your local devices to provide more options.

You could maybe use EIGRP stub feature and Summary addresses to allow default routes from the remote devices.

ip summary-address eigrp <AS#> 0.0.0.0 0.0.0.0

Put that on each remote interface towards your devices. The will advertise a default route to its EIGRP neighbor. You can modify interface metrics so it chooses one over the other. Then you can advertise that default route into the rest of your EIGRP Autonomous System.

[TRACY HARTMANN]

How does the ip summary-address eigrp command work.   On the core switch that has the circuits on the remote end would I add this statement on the interfaces?  I don't want to advertise a default route.

[MHM Cisco World]

you have fiber do you run UDLD ?

[TRACY HARTMANN]

No we do not run UDLD.  Thanks 

[paul driver]

Hello
If you have two default static routes unless you make them conditional, the situation you have described can be incurred as such the next-hop isn’t of that failed static route isnt reachable
however there rtr still sees it a valid thus blackholing your traffic that need to be forwarded on the wan

example ISLA-tracking:
ip sla 1
icmp-echo 8.8.8.8 source-interface xx
timeout 1000
threshold 2000
frequency 15

ip sla schedule 1 start now life forever
track 10 rtr1 reachabilty

access-list 100  permit icmp host <source ip> host 8.8.8.8 echo

route-map ipsla
match ip address 100
set ip next-hop (next hop of interface fof wan port)
set interface Null0

ip local policy route-map ipsla
ip route 0.0.0.0 0.0.0.0  <primary interface> <nexthop ip> track 10 name primary link
ip route 0.0.0.0 0.0.0.0 <secondary interface> <nexthop ip> 2 name backup link.